linux-kernel - Re: [PATCH 04/14] KVM: VMX: Extend VMX controls macro shenanigans

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YrpbiWw1E4DXQ962@google.com>
Date:   Tue, 28 Jun 2022 01:38:17 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     "Dong, Eddie" <eddie.dong@...el.com>
Cc:     Vitaly Kuznetsov <vkuznets@...hat.com>,
        "kvm@...r.kernel.org" <kvm@...r.kernel.org>,
        Paolo Bonzini <pbonzini@...hat.com>,
        Anirudh Rayabharam <anrayabh@...ux.microsoft.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Maxim Levitsky <mlevitsk@...hat.com>,
        "linux-hyperv@...r.kernel.org" <linux-hyperv@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 04/14] KVM: VMX: Extend VMX controls macro shenanigans

On Mon, Jun 27, 2022, Dong, Eddie wrote:
> >  static inline void lname##_controls_clearbit(struct vcpu_vmx *vmx, u##bits
> > val)	\
> >  {
> > 	\
> > +	BUILD_BUG_ON(!(val & (KVM_REQ_VMX_##uname |
> > KVM_OPT_VMX_##uname)));	\
> >  	lname##_controls_set(vmx, lname##_controls_get(vmx) & ~val);
> > 	\
> >  }
> 
> With this, will it be safer if we present L1 CTRL MSRs with the bits KVM
> really uses? Do I miss something?

KVM will still allow L1 to use features/controls that KVM itself doesn't use, but
exposing features/controls that KVM doesn't use will require a more explicit
"override" of sorts, e.g. to prevent advertising features that are supported in
hardware, known to KVM, but disabled for whatever reason, e.g. a CPU bug, eVMCS
incompatibility, module param, etc...

The intent of this BUILD_BUG_ON() is to detect KVM usage of bits that aren't enabled
by default, i.e. to lower the probability that a control gets used by KVM but isn't
exposed to L1 because it's a dynamically enabled control.